Mar

12

 Hi Vic,

Sorry I didn't get to say hello at the last Junto meeting, which was terrific.

Regarding the question you asked at the event: This is not a comment on the validity of Mackey or Teicholz's position. I think Gene was right about needing to do the marathon rather than the sprint, i.e., reading up in greater detail to get fuller knowledge of the topic. But you asked Teicholz whether she rejected the findings of epidemiological studies in the face of many experts who considered them valid.

My question is how you come down on The Phenomenon Formerly Known as Global Warming, i.e., climate change, and whether the skeptics (called "deniers," as in "Holocaust deniers," by the proponents) should bow to the opinion of the majority of experts, notwithstanding the fact that the 30-40 years ago the experts were very concerned about Global Cooling.

I think you or one of the other questioners raised the point that medical journals have 20 referees on an article. So perhaps the studies are more reliable than those in financial journals, where a professor at UMass assigns his grad students to study the results reported in articles and finds that 25% have serious errors. No need to write a treatise on this, but I'm interested in your thoughts.

Best regards,

Marty

Victor Niederhoffer replies:

The hazard rate is the gold standard in epidemiological studies. And the hazard rate for life expectancy in the typical studies of meat versus vegetable studies is about 2 to 1. You can imagine the significance of such a difference with 80,0000 subjects or so. The estimate of p has a binomial distribution. When you have a sample in the millions as the meta studies of such divergences have, there is no room for selective sampling. What are the chances that a million people are selectively different from the remained. The statistics of quota sampling are relevant here.

I would also point out that all the epidem studies use the cox hazards model as their statistical foundation. The cox model is like a standard multiple regression model but deals with probabilities instead of levels or changes. All such studies control for all measurable independent variables such as health status, smoking history , and weight. The chances that other independent variables mite affect the outcome with numbers this large in the meta studies is zero.

 You ask why science is sometimes wrong about such things, and I would say that the epidem studies with millions of subjects in many different counties with many different independent variable and selected groups are in a completely different kind of scientific category than climate change where the dependent variables are temperature changes over time, and the independent variables are chemical entities with say 10 or 20 observations and more independents than observations. There are no epidem studies possible in climate change.

I believe Nina is very out of her area of competence and it was dysfunctional to have her armchair rebuttals of highly refereed studies and statistically significant results in the one in a billion categories based on sampling errors and hypothetical differences in the groups studied.

Victor Niederhoffer adds:

 Another way of responding to this tomfoolery less statistically and more qualitatively and common sense is this. The differences between the groups that eat meat and eat vegetables in the many life expectancy studies are of the order of 5 to 10 years. Such differences are important to most human beings who wish to live. With samples in the millions in aggregate and attempts to control for all measurable independent variables there is a reasonable likelihood that the two groups are representative of the populations.

Now if there was a difference in these groups with respect to some other variable, like exercise or preference for risky activities, and somehow that was not covered by the other independent variables, then what could that variable be. Let us find it, because it would be the key to health. i.e. if it had so much of an impact by not being measured, it is truly important, much more important than the diet, and if it could be found it would be the open sesame.

But of course it can't be found because it would be statistically impossible to find something uncorrelated with the other variables that have been studies that has such an effect. It only exists in the armchair and the debunking retrospection and Monday morning quarterbacking. 


Comments

WordPress database error: [Table './dailyspeculations_com_@002d_dailywordpress/wp_comments' is marked as crashed and last (automatic?) repair failed]
SELECT * FROM wp_comments WHERE comment_post_ID = '10140' AND comment_approved = '1' ORDER BY comment_date

Name

Email

Website

Speak your mind

Archives

Resources & Links

Search